Journal of Gastroenterology and Hepatology (2001) 16, 1254–1259

after alcohol withdrawal. However, there have been fewstudies addressing hepatic biochemical changes duringalcohol withdrawal in subjects across the full spectrumof alcoholic liver disease. These studies were often retrospective, and their results were conflicting.3–6

There are several mechanisms by which abrupt with-drawal might cause liver damage. Ischemia may result

INTRODUCTION

In clinical practice, we have observed that some alco-holic patients deteriorate both clinically and biochemi-cally over a few days following withdrawal from alcohol.Case reports and studies of alcoholic hepatitis1,2 alsosuggest biochemical liver function tests may worsen

LIVER FIBROSIS: EFFECT OF ANTI-OXIDANT,ALCOHOL WITHDRAWAL

Effect of alcohol withdrawal on liver transaminase levels andmarkers of liver fibrosis

STEWART CAMPBELL,*,§ PETER M TIMMS,† PAUL R MAXWELL,†

ELIZABETH M DOHERTY,‡ MOHAMMAD Z RAHMAN,‡ MICHAEL EJ LEAN§ AND BOOTH J DANESH*

Departments of *Gastroenterology and †Biochemistry, Stobhill Hospital, Department of ‡Psychiatry, RuchillHospital and Department of §Human Nutrition, Glasgow Royal Infirmary,Glasgow, Scotland, United

Kingdom

AbstractBackground and Aim: Acute alcohol withdrawal causes changes in hepatic blood flow and metabo-lism that may result in liver damage.This study aims to assess liver function tests and markers of hepaticfibrogenesis following alcohol withdrawal in alcoholics with clinically compensated liver disease.Methods: Serial liver function tests and clinical assessments were performed on 22 male alcoholicsduring alcohol withdrawal. Plasma tissue inhibitor of metalloproteinase 1 (TIMP1), an inhibitor of col-lagen degradation, and plasma amino-terminal procollagen III peptide (PIIINP), a collagen precursormolecule, were measured in these alcoholics and in 11 control subjects.Results: Transaminase levels did not change significantly over 7 days when all subjects were analyzedtogether. However, 32% of subjects showed a marked transaminase rise. These subjects did not differfrom the others in baseline characteristics or short-term outcome, but had a greater benzodiazepinerequirement. Only one subject consumed paracetamol (acetaminophen; 1–2 g/day). He had the largesttransaminase rise. By comparing PIIINP assays, intact PIIINP concentration appears to increase fol-lowing alcohol withdrawal. The TIMP1 levels were elevated in alcoholic subjects, but did not changefollowing withdrawal.Conclusions: Increasing PIIINP suggests that hepatic fibrogenesis increases, or hepatic clearance falls,during acute alcohol withdrawal. The TIMP1 elevation in these alcoholics suggests that the inhibitionof collagen degradation occurs while liver disease is still compensated. The period following alcoholwithdrawal may be a time of marked increased susceptibility to paracetamol. The biochemical changeswe observed were not associated with adverse short-term outcome, but the cumulative effect afterrepeated episodes of abrupt withdrawal may be of concern.© 2001 Blackwell Science Asia Pty Ltd

Key words: acetaminophen, alcohol withdrawal, alcoholic liver disease, liver transaminases, procolla-gen peptide, tissue inhibitor of metalloproteinase 1.

Correspondence: Dr S Campbell, 32 Dalziel Drive, Pollokshields, Glasgow G41 4HY, Scotland, United Kingdom.Email: gcl028@clinmed.gla.ac.uk

Accepted for publication 15 August 2001.

Hepatic biochemistry after alcohol withdrawal 1255

from a decrease in portal blood flow and increasedhepatic oxygen demand upon cessation of alcoholintake.7 Chronic alcohol abuse induces microsomalliver enzymes that are capable of generating toxicmetabolites. During alcohol withdrawal, these enzymesare no longer saturated by ethanol and hence areincreasingly available to metabolize drugs or other sub-stances to potential hepatotoxins.8

Studies of fibrotic markers have shown an increase incollagen metabolites following alcohol withdrawal.However, the conclusions of these studies are unclearbecause of difficulties in interpreting the results with thetype of assay used.9

Our aim was to determine the pattern of change ofstandard biochemical tests in alcohol withdrawal, andto use non-invasive markers of liver disease to look forfurther evidence of hepatic damage.

METHODS

Patients

Twenty-two male patients (median age 44 years, range29–65), all of whom had consumed > 150 g of alcoholdaily for > 1 year (median 10 years) were recruited fromadmissions to the alcohol rehabilitation unit at RuchillHospital, Glasgow. Admissions were asked consecu-tively to participate in the present study, provided theywere not excluded by the criteria below. Patients wereadmitted for elective alcohol withdrawal, rather thanmanagement of liver disease. We chose a patient popu-lation with clinically compensated liver disease. Acuteillness often precipitates alcohol withdrawal in chronicalcoholics, and could potentially confound our re-sults. We therefore excluded subjects with evidence of a medical condition precipitating alcohol withdrawal or who had hepatic decompensation (jaundice, as-cites, encephalopathy, or gastrointestinal hemorrhage)at baseline. Patients with evidence of liver disease ofanother etiology (positive hepatitis A, B, or C serology,positive antimitochondrial, antismooth muscle or anti-nuclear antibody, or a history of gallstones) wereexcluded from the present study. These subjects weremanaged as inpatients for 1 week following alcoholwithdrawal, and they were asked to return as out-patients during the following week. They were with-drawn from the study if they resumed drinking. Allsubjects gave informed written consent and the localethics committee approved this study. This study con-forms to the ethical guidelines of the 1975 Declarationof Helsinki.

Clinical measurements

Patients were examined daily for signs of liver diseaseand possible complications such as infection, gastroin-testinal bleeding, ascites, edema or pancreatitis. An esti-mate of body composition was made on admission bythe measurement of body mass index (BMI), tricepsskin-fold thickness and mid upper arm circumference.

The percentage body fat was calculated from thesemeasurements.10 In addition, daily measurements ofalcohol withdrawal symptoms, for example, tremor, hal-lucination, were made by using the Clinical InstituteWithdrawal Assessment Scale for Alcohol (revised;CIWA-Ar).11 Alcohol intake was assessed on the basisof clinical history obtained by medical and nursing staffexperienced in alcohol dependence.

Laboratory measurements

The first blood samples were obtained on hospitaladmission within 72 h of the last alcohol consumed(median 22 h).These samples were processed as part ofthe routine hospital assays for the following: alaninetransaminase (ALT), albumin, alkaline phosphatase(ALP), amylase, aspartate transaminase (AST), biliru-bin (BR), blood ethanol, coagulation profile, creatinekinase (CK), full blood count, gamma glutamyl transferase (GGT), urea and electrolytes. These wererepeated daily for 7 days.The patients were usually dis-charged at day 7, and returned on day 14 to repeat thesetests. We arbitrarily defined worsening of liver functiontests as a 50% rise from the baseline value, with the ele-vated result above the normal laboratory referencerange. This was chosen prior to the commencement ofthe present study in an attempt to exclude values thatrose simply as a result of minor biological and analyti-cal variation. Continuing abstinence was assessed byblood alcohol measurement, random breath alcoholmeasurement, and self-reporting.

Plasma samples from days 1, 7 and 14 were frozenand later analyzed in one batch, to minimize analyticalvariation, for amino-terminal procollagen III peptide(PIIINP) by using commercial radio-immunoassay kitsfrom Orion Diagnostica, Espoo, Finland and Cis BioInternational, Paris, France, and for tissue inhibitor of metalloproteinase-1 (TIMP-1) by using a com-mercial colorimetric ELISA assay from Amersham Life Science, Buckinghamshire, England. Two differentassays were used for PIIINP. The Orion assay, whichuses an antibody to human PIIINP, is specific for theintact procollagen peptide in human serum,12 and isregarded as a marker for collagen synthesis.13 The Cisassay uses an antibody to bovine PIIINP and it is sen-sitive for both intact PIIINP and smaller PIIINP frag-ments in human serum.14 It is regarded as a marker forboth collagen synthesis and degradation.13 Our intra-assay coefficients of variation for these assays were:TIMP1, 2.9%; PIIINP (Orion), 4.8%; and PIIINP(Cis), 4.8% at levels measured in the present study.

Plasma from 11 healthy age-matched males who hadno history of excess alcohol consumption or physicalillness was used to provide a normal range for PIIINPand TIMP1.

Statistical analysis

In the comparison of baseline data, unpaired normaldata were analyzed by using a Student’s t-test,

1256 S Campbell et al.

non-parametric data were analyzed by using theMann–Whitney U-test. An estimate of 95% confidenceinterval for prevalence was performed by the methoddescribed by Bland.15 For the analysis of repeated measures data, within-subjects analysis of variance wasused. All tests were two-tailed, and statistical signifi-cance was taken as P < 0.05. Statistical analyses wereperformed by using SPSS (version 9.0; SPSS Incorpo-rated, Chicago, IL, USA) and MINITAB (version 11;Minitab Incorporated, Philadelphia, PA, USA) statisti-cal software.

RESULTS

Data were analyzed on an intention-to-include basis.Twenty-one (95%) of the subjects remained in thepresent study for the first week to complete the initialintensive assessment phase, and 13 subjects (59%)remained until Day 14.

Biochemical liver function tests

Gamma GT, bilirubin, ALP, and CK all tended toimprove or remain the same in the 7 days followingalcohol withdrawal (data not shown). Aspartate amino-transferase and ALT also tended to improve in themajority of subjects (Fig. 1), and there was no signifi-cant change in transaminases when the group was considered as a whole, but a number of subjects didshow a marked worsening in liver transaminase levels(Fig. 2).

The change in transaminases was of similar propor-tion for AST and ALT. By using our predefined cri-teria described above, we separated the subjects into two groups, shown graphically as an improv-ing transaminase group (Fig. 1) and as a worseningtransaminase group (Fig. 2). Seven out of 22 subjectsenrolled showed a worsening in transaminases (32%;

95% confidence interval, 12–51%). Despite a rise inALT, subject A in Fig. 1 was included in the improvingtransaminase group as he did not meet the predefinedcriteria for the other group.

There was no statistically significant differencebetween the groups with respect to the biochemicalparameters listed above or to the following clinical para-meters: age, daily alcohol consumption, years of alcoholabuse, cigarettes smoked per day, percentage body fat,duration of abstinence prior to admission. The normalor mildly elevated baseline biochemical results for bothgroups are consistent with our intention to recruit sub-jects with clinically compensated liver disease.

The group with worsening transaminases receivedsignificantly greater amounts of chlordiazepoxide thanthe other group (median 140 mg/day) during the firstweek of admission compared to the remaining subjects(median 36 mg/day), P = 0.003. The CIWA scores werenot different among the two groups (median 16, wors-ening transaminase group, vs 12, improving transami-nase group, P = 0.3).

Only one subject received paracetamol-containingmedication (subject B in Fig. 2). He took a total of1 g paracetamol on days 3, 4, 6 and 8, and 2 g on day7. He had the most marked transaminase rise of anysubject in this study.

The rate of complications during admission wassimilar in the two groups, all of which were respiratoryinfections (one/seven in the worse transaminase group,three/17 improving transaminase group, P > 0.9).

Markers of hepatic fibrogenesis

The TIMP1 levels were markedly higher in the alcoholicgroup (P < 0.001). The PIIINP concentrations from the alcoholic patients were significantly lower than incontrol subjects in the Cis assay (P = 0.03), but not inthe Orion assay (P = 0.96), implying that the differencebetween these groups is predominantly because of thePIIINP fragments (Fig. 3).

Figure 1 Change in serum alanine aminotransferase (ALT)following alcohol withdrawal: improving transaminase group(ALT reference range 5–40). Subject A showed a transami-nase rise but did not meet predefined criteria for inclusion inthe worsening transaminase group.

450

400

350

300

250

200

150

100

50

00

ALT

Subject A

1 2 3 4 5 6 7 8

Time after admission for alcohol withdrawal (days)

450

400

350

300

250

200

150

100

50

0

ALT

0 1 32 4 5 6 7 8

Subject B

Time after admission for alcohol withdrawal (days)

Figure 2 Change in serum alanine aminotransferase (ALT)following alcohol withdrawal: worsening transaminase group.(ALT reference range 5–40). Subject B was the only subjectto have taken paracetamol.

Hepatic biochemistry after alcohol withdrawal 1257

There was an increase in the PIIINP levels followingalcohol withdrawal in both assays (Orion P < 0.02, CisP < 0.01). The percentage median increase was: Orion44 and 75%, Cis 10 and 42% at days 7 and 14, respec-tively. As the major rise was seen with the Orion assay,this would be consistent with a rise mainly caused byintact PIIINP. The TIMP1 levels did not change fol-lowing alcohol withdrawal (P = 0.22; Fig. 3).

There was no significant difference between thegroup with improving transaminases and the otherswith respect to day 1 TIMP1 or PIIINP measured byeither assay (Cis median 0.36 vs 0.46, P = 0.64; Orionmedian 2.82 vs 2.44, P = 0.74; TIMP1 mean 1208 vs1268, P = 0.9 in worsening and improving transaminasegroups, respectively).

DISCUSSION

The pathophysiology of hepatic damage in alcoholismis not fully understood. However, there is evidence formechanisms by which hepatic damage may occur inalcohol withdrawal. Portal blood flow increases follow-ing alcohol ingestion,16 and returns to previous levelsupon alcohol withdrawal. The hepatic oxygen demandincreases after alcohol consumption, and remainsincreased for several days after alcohol withdrawal.7

This may result in hypoxic hepatic damage. Chronicalcohol abuse induces the hepatic cytochrome P450oxidase enzymes.While drinking alcohol, these enzymesare saturated by ethanol. However, upon alcohol with-drawal these enzymes are free to metabolize other sub-strates, including drugs such as paracetamol, topotentially hepatotoxic metabolites.8

The final outcome of long-term liver damage in alco-holic liver disease is cirrhosis. A large proportion of thecollagen deposited, especially in the early stages of cir-rhosis, is type III.17 The precursor molecule of type IIIcollagen, type III procollagen, is secreted with peptideextensions called propeptides (PIIIP) attached to theamino (PIIINP) and carboxy (PIIICP) terminals. IntactPIIINP is released when procollagen molecules aggre-gate to form the type III collagen fiber.18 However, somePIIINP is not released from collagen after the collagenfiber has been formed. Type III collagen fibers withPIIINP still attached have been demonstrated inprimate liver.19 If collagen to which PIIINP remainsattached is subsequently degraded, then smaller frag-ments of PIIINP may be released.Thus, a PIIINP assaythat is sensitive to both intact PIIINP and PIIINP

1.2

1.1

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

PIII

NP

(C

is)

U/m

L

(a)P < 0.05 P < 0.01

P < 0.001

P < 0.02P = NS

P = NS

16.0

14.0

12.0

10.0

8.0

6.0

4.0

2.0

0.0

PIII

NP

(or

ion)

mg/

L

(b)

(c)

3000

2500

2000

1000

1500

500

0

TIM

P1

ng/m

L

Control Day 1 Day 7 Day 14

Controls and alcoholics (days after admission)

{ {} }

{ {} }

{ {} }

Figure 3 Boxplot showing plasma concentrations of (a)plasma amino-terminal procollagen III peptide (PIIINP; Cis),(b) PIIINP (Orion) and (c) plasma tissue inhibitor of metal-loproteinase 1 (TIMP1) in controls and alcoholics undergo-ing alcohol withdrawal. Boxes represent the interquartilerange. Horizontal bar represents the median. (�) Representsoutlying values. Statistical significance of the differencebetween alcohol withdrawal subjects and control subjects isindicated. NS, non-significant.

�

1258 S Campbell et al.

fragments reflects both synthesis and degradation (theCis assay).13,20 Assays measuring intact PIIINP only(the Orion assay) reflect mainly collagen synthesis.12,13

Plasma PIIINP levels correlate well with hepatic prolylhydroxylase levels.21 Direct mathematical comparisonof the two assays was not possible, but by comparingthe pattern of changes in these two assays, we hope todetermine whether a change in PIIINP is a result ofchanges in intact PIIINP or PIIINP fragments. Therewas no assay available to measure PIIINP fragmentsalone.

The PIIINP is metabolized by hepatic sinusoidalcells,22 and serum levels may therefore be affected bychanges in clearance resulting from a decrease inhepatic blood flow during alcohol withdrawal.

Failure to degrade excess collagen is important in thedevelopment of hepatic cirrhosis. A family of metallo-proteinase enzymes is largely responsible for degradingnative collagen. The TIMP1 is secreted by hepatic stel-late cells and reduces markedly the activity of metallo-proteinases and, therefore, decreases the degradation of extracellular matrix.23 Hepatic collagenase activitydecreases as hepatic fibrosis progresses.24

There is good evidence that plasma levels of TIMP1give an accurate reflection of liver collagen metabolism,when other serious systemic diseases can be excluded.Plasma TIMP1 levels correlate with fibrosis scores onliver biopsy.25,26 Plasma TIMP1 (and PIIINP) concen-trations rise as alcoholic liver disease progresses in his-tological severity.27–29 Plasma TIMP1 levels are notknown to be influenced by hepatic blood flow.

Although there was no overall change in transami-nases following alcohol withdrawal when considering all the subjects, there was a subset showing a markedtransaminase increase. The subject with the largesttransaminase rise was the only subject to consumeparacetamol, albeit at modest doses. The potential forparacetamol-induced hepatotoxicity8 might therefore bemarkedly increased, being manifest even if a dose wellbelow the recommended daily maximum dose isingested at a critical time following alcohol withdrawal.

A large retrospective Scandinavian study6 showed anincrease in AST and ALT of 8.7 and 7.0%, respectively,of subjects withdrawing from alcohol. However, liverfunction tests were only checked on the basis of clini-cal need, and in most subjects these were not checkedon more than two occasions during the first week ofwithdrawal. They found that an elevation in transami-nases was associated with more severe symptoms ofdelirium tremens, which is supported by our finding ofa greater requirement for benzodiazepines. We suggestthat the worsening transaminase group had a moresevere alcohol withdrawal despite similar CIWA scoresbecause they required more benzodiazepine, which willtend to reduce their CIWA score. Sympathetic over-activity contributes to the symptoms observed inalcohol withdrawal and can further increase hepaticoxygen demand,30 and hence those patients with moresevere symptoms may also have a greater degree ofhepatic ischemia resulting in altered transaminases.

A retrospective study of liver function tests and liverbiopsy findings in alcohol withdrawal suggested thatthose with worsening liver function tests might have a

histologically more severe disease.5 The retrospectivenature of that study may have introduced bias.We foundno difference in biochemical or clinical parameters tosuggest that our subgroup with worsening transami-nases had a more severe disease. It was thought to beunethical to perform a liver biopsy in our subjects whohad clinically compensated disease.

Our control subjects appeared to have a higher levelof PIIINP fragment than the alcoholic subjects onday 1.This observed difference may have resulted froma decreased collagen degradation in alcoholics whilethey were still drinking, which would be consistent withthe findings of increased serum TIMP in alcoholics.Lieber’s group also noted a higher PIIINP level in con-trols versus alcoholics with fatty liver when using anassay that was very sensitive for PIIINP fragments, butnot when using a different assay which was less sensi-tive.27–29 This group has also shown that alcoholics havean increase in PIIINP following alcohol withdrawal.27,29

However, these studies did not employ an assay exclu-sively for intact PIIINP, and we were hence unable toconclude if the observed changes were likely to bebecause of the intact peptide or peptide fragments.

We demonstrated a rise in PIIINP following alcoholwithdrawal that is likely to be caused by intact PIIINP.Although this may imply increased fibrogenesis, it may reflect decreased hepatic clearance. The markedincrease in TIMP1 in alcoholics compared to controlssuggests that there may be significant inhibition of col-lagen degradation even in the group of patients with aclinically compensated disease.

CONCLUSIONS

Our results suggest that in alcoholics with a clinicallycompensated liver disease, alcohol withdrawal is usuallyaccompanied by an improvement in liver transaminaselevels. However, a minority of subjects showed a wors-ening of transaminases. The period following alcoholwithdrawal may be a time of marked increased susceptibility to the effects of even small amounts of paracetamol. Although the changes observed intransaminases and PIIINP are small and not associatedwith major differences in a short-term clinical outcome,they may reflect a mild liver injury and either a minorincrease in fibrosis or a decrease in hepatic clearance.The cumulative effect of repeated episodes of with-drawal may be important, and we suggest that furtherstudies looking at the effect of gradual withdrawal ofalcohol on clinical and biochemical parameters shouldbe performed. The most striking observation is themarked increase of serum TIMP1 levels in these alcoholics compared with controls, supported by thefinding of decreased PIIINP fragments, suggesting thatdecreased collagen degradation may be present prior tothe development of overt liver disease.

REFERENCES

1 Sabesin SM, Hawkins HL, Bertram PD, Mann JA, PeaceRJ. Alcoholic hepatitis. Gastroenterology 1978; 74: 276–86.

Hepatic biochemistry after alcohol withdrawal 1259

2 Helman RA, Temko MH, Nye SW, Fallon HJ. Alcoholichepatitis: natural history and evaluation of prednisolonetherapy. Ann. Intern. Med. 1971; 74: 311–21.

3 Hemmingsten R, Kramp P, Dissing J. Delirium Tremens:Some clinico-chemical features. Acta Psychiat. Scand.1980; 62: 503–10.

4 Nielsen J. Delirium tremens in Copenhagen. Acta Psychiat. Scand. 1965; 41 (Suppl. 187): 60–92.

5 Marshall JB, Burnett DA, Zetterman RK, Sorrell MF.Clinical and biochemical course of alcoholic liver diseasefollowing sudden discontinuation of alcoholic consump-tion. Alcohol Clin. Exp. Res. 1983; 7: 312–15.

6 Salum I. Delirium tremens and certain other acute sequelsof alcohol abuse. Acta. Psychiat. Scand. Supplement 1972;235: 1–119.

7 Orrego H, Carmichael FJ. Effects of alcohol on liverhaemodynamics in the presence and absence of liverdisease. J. Gastroenterol. Hepatol. 1992; 7: 70–89.

8 Lieber CS. Biochemical factors in alcoholic liver disease.Semin. Liver Dis. 1993; 13: 136–53.

9 Risteli J, Risteli L. Analysing connective tissue metabolitesin human serum. Biochemical, physiological and method-ological aspects. J. Hepatol. 1995; 22 (Suppl. 2): 77–81.

10 Lean MEJ, Han TS, Deurenberg P. Predicting body com-position by densitometry from simple anthropometricmeasurements. Am. J. Clin. Nutr. 1996; 63: 4–14.

11 Sullivan JT, Sykora K, Schneiderman J, Narango CA,Sellers EM. Assessment of alcohol withdrawal: the revisedClinical Institute Withdrawal Assessment for Alcohol scale(CIWA-Ar). Br. J. Addiction 1989; 84: 1353–7.

12 Risteli J, Niemi S, Trivedi P, Maentausta O, Mowat AP,Risteli L. Rapid equilibrium radioimmunoassay for theamino-terminal propeptide of human type III procollagen.Clin. Chem. 1988; 34: 715–18.

13 Schuppan D, Stolzel U, Oesterling C, Somasundaram R.Serum assays for liver fibrosis. J. Hepatol. 1995; 22 (Suppl.2): 82–8.

14 Rohde H, Vargas L, Hahn E, Kalbfleish H, Bruguera M,Timpl R. Radioimmunoassay for type III procollagenpeptide and its application to human liver disease. Eur. J.Clin. Invest. 1979; 9: 451–9.

15 Bland JM. An Introduction to Medical Statistics, 2nd edn.Oxford: Oxford University Press, 1995.

16 Piano MR, Holm K, Melchior CL, Ferguson JL.Alteration in splanchnic blood flow after low and highdoses of ethanol. Alcohol Clin. Exp. Res. 1991; 15: 106–11.

17 Popper H, Udenfriend S. Hepatic fibrosis-correlation ofbiochemical and morphologic investigations. Am. J. Med.1970; 49: 707–21.

18 Prockop DJ, Kivirikko KI,Tuderman L, Guzman NA.Thebiosynthesis of collagen and its disorders (first of twoparts). N. Engl. J. Med. 1979; 301: 13–23.

19 Sato S, Leo MA, Lieber CS. Ultrastructural localisationof type III procollagen in baboon liver. Am.J.Pathol. 1986;122: 212–17.

20 Hahn EG. Blood analysis for liver fibrosis. J. Hepatol.1984; 1: 67–73.

21 Torres-Salinas M, Pares A, Caballeria J et al. Serum pro-collagen type III peptide as a marker of hepatic fibro-genesis in alcoholic hepatitis. Gastroenterology 1986; 90:1241–6.

22 Smedsrod B. Aminoterminal propeptide of type III pro-collagen is cleared from the circulation by receptor medi-ated endocytosis in liver endothelial cells. Collagen. Rel.Res. 1988; 8: 375–88.

23 Iredale JP, Murphy G, Hembry RM, Freidman SL, ArthurMJP. Human hepatic lipocytes synthesize tissue inhibitorof metalloproteinases-1 (TIMP1): implications for regu-lation of matrix degradation in the liver. J. Clin. Invest.1992; 90: 282–7.

24 Maruyama K, Feinman L, Fainsilber Z, Nakano M,Okazaki I, Lieber CS. Mammalian collagenase increasesin early alcoholic liver disease and decreases with cirrho-sis. Life Sci. 1982; 30: 1379–84.

25 Murawaki Y,Yamamoto H, Kawasaki H, Shima H. Serumtissue inhibitors of metalloproteinases in patients withchronic liver disease and with hepatocellular carcinoma.Clin. Chim. Acta 1993; 218: 47–58.

26 Murawaki Y, Ikuta Y, Idobe Y, Kitamura Y, Kawasaki H.Tissue inhibitor of metalloproteinase-1 in the liver ofpatients with chronic liver disease. J. Hepatol. 1997; 26:1213–19.

27 Li J, Rosman AS, Leo MA, Nagai Y, Lieber CS. Tissueinhibitor of metalloproteinase is increased in the serum of precirrhotic and cirrhotic alcoholic patients and canserve as a marker of fibrosis. Hepatology 1994; 19:1418–23.

28 Sato S, Nouchi T, Worner TM, Lieber CS. Liver fibrosisin alcoholics. Detection by fab radioimmunoassay ofserum procollagen III peptides. JAMA 1986; 256: 1471–3.

29 Nouchi T,Worner TM, Sato S, Lieber CS. Serum procol-lagen type III N-terminal peptides and laminin P1 peptidein alcoholic liver disease. Alcohol Clin. Exp. Res. 1987; 11:287–91.

30 Hadengue A, Moreau R, Lee SS, Gaudin C, Rueff B, Lebrec D. Liver hypermetabolism during alcohol withdrawal in humans. Gastroenterology 1988; 94:1047–52.